INSPIRE cfMeDIP-seq Analysis
In the interest of transparency, openness, and reproducibility, this report contains the analysis code which generates all of the figures and numerical results for the manuscript entitled Early changes in tumor-agnostic cell-free methylomes and fragmentomes predict outcomes in patients with solid tumors treated with pembrolizumab. In it, we report on the results of cell-free methylated DNA immunoprecipitation and sequencing (cfMeDIP-seq) from the investigator-initiated phase 2 study of pembrolizumab immunological response evaluation (INSPIRE).
Using Rstudio or the knitr package in R, all results reported in the
manuscript can be regenerated in full using this RMarkdown file. All of
the processed, deidentified data are stored in
figure_data.Rds as dataframes and are attached to the
environment in the setup block. Our nomenclature is variable names
starting with df to denote source data tables and
plotdata for processed dataframes used as intermediates to
generate figures. Below is the high-level description of source
dataframe.
df_samples: Mapping of technical sequencing library IDs to patient/timepoint IDs.df_clinical: Deidentified clinical data, with emphasis on RECIST best response and survival outcomes.df_methylation_signature: All sites in the methylation signature and associated metadata.df_posterior: Complete table of methylation posterior probabilities, one for each sample and genomic window within the methylation signature.df_csm: Cancer-specific methylation scores, one for each sample of INSPIRE patients.df_csm_normalcontrol: Cancer-specific methylation scores, one for each sample of normal controls.df_natera_response: Bespoke mutation-based ctDNA levels per patient, with change from baseline to cycle 3. Same values as reported by Bratman & Yang et al. 2020.df_natera_timepoint: Bespoke mutation-based ctDNA levels, one per sample.df_biomarkers: Immunologic biomarkers used in multivariate analysis, specifically PD-L1 expression and tumor mutation burden.df_fls: Fragment length scores, one per sample.df_fragment_ratio: Fragment length ratios in 5 Mb segments.df_fragment_ratio_vartest_regions: Results of variance tests for fragment length ratios of 5 Mb segments.df_insertsize: Global fragment length distributions extracted by Picard.df_nucleosomes: Nucleosome position features.df_example_coverage: cfMeDIP-seq window coverage within the CSM methylation signature for nine example patients for illustrating how CSM is calculated.df_clearance_natera: Bespoke mutation-based ctDNA clearance status, as reported by Bratman & Yang et al. 2020.df_chromosome_lengths: hg38 chromosome lengths.df_read_counts: Actual read counts for each sequenced library.